Unstructured input becomes structured value. Committees of models keep the answers honest. Domain workflows turn both into tools experts rely on.
Days of manual work → minutes. Source-attributed at every stage.
Work with us →HITS systems run on three layers. Ingestion engines turn messy input — PDFs, images, audio, scans, domain-specific documents — into structured, source-attributed data. Agent committees turn the leading answer into the answer that survived structured attempts to break it. Workflow products turn both into tools domain experts rely on. We pick whichever stack fits the job — Tauri, React Native, Next.js, FastAPI, polyglot persistence. The sophistication isn't in any single layer — it's in the orchestration.
Route by task difficulty — smallest model that works. Prompt-cached. Typically ~60% cost reduction versus single-tier.
The schema is the contract. Extracted fields have operator, value, unit — actionable downstream, not just readable.
Every extracted fact links back to its source document and page. Non-negotiable.
Domain experts (estimators, research nurses, genealogists) validate and refine. The model does the volume; humans keep the ground truth.
Three flagship products. Same pattern, different domains.
Local-first desktop genealogy and family legacy platform.
Build family trees, attach media, narrate stories in your ancestors’ voices, run AI research agents 24/7. Local-first. Users own every byte.
Genealogists, family historians · Local-first, user-owned · In development
More ▸ (click to collapse)Desktop app with an embedded local store and vector index, paired with cloud APIs for subscription-gated AI features. AES-256 at rest, collaborative tree merges, and a full export path. The user keeps everything.
Clinical trial protocol extraction, plus a free Trial Finder for oncologists.
Ingests 100–400 page oncology protocols and emits structured eligibility rules, visit schedules, and workflow documents. Every artifact cross-validated against the source.
Clinical research teams, oncologists · Domain-expert validated · In development
More ▸ (click to collapse)B2B SaaS for clinical research teams plus a free patient-facing Trial Finder — both powered by the same eligibility extraction engine. Screen-failure analytics are where the real ROI lives.
Committee-based AI for high-stakes specialist work.
Routes consequential decisions to a committee of models from different families, runs adversarial agents against the output, and emits a verdict only when the committee converges. Disagreement is signal — kept, not hidden.
AI teams shipping production work · Regulated and specialist domains · In development
More ▸ (click to collapse)Multi-model extraction with adversarial cross-validation. Every answer is challenged before it ships; convergence between independent re-runs is the gate. Built for the work where wrong answers carry real cost — clinical eligibility, legal review, compliance, domain-expert workflows.
Ten principles that shape every HITS product.
Route by task difficulty: smallest model that works for classification, mid-tier for extraction, largest only for validation. Cache system prompts. Never pay for reasoning you don’t need.
The schema is the contract. If a field isn’t structured — operator, value, unit — it isn’t actionable downstream. Design schemas with the end workflow in mind.
Single-pass extraction is a demo. Production systems dispatch adversaries to break their own output, and accept only on convergence.
Every extracted fact links back to its source document and page. Non-negotiable.
Failures happen. Systems resume from the last completed stage, not from scratch. State machines with atomic transitions.
Pick the right datastore for each workload — relational for truth, vector for semantic search, graph for relationships, local for local-first. Don’t unify for unification’s sake.
Data mutations emit events; independent consumers handle side effects without coupling.
For consumer software, the user owns the data. Cloud extends capability, not ownership.
Shared patterns across products are surfaced explicitly so sibling products don’t reinvent.
Ground truth from domain experts replicates slowly; code replicates fast. Build where domain expertise compounds.
Agent committees for the decisions that matter.
When a HITS engine reaches a decision worth doing right, it doesn't ask one model. It convenes a committee.
Different model families. Different priors. Different prompts. Each member produces a candidate answer with cited evidence. Candidates are scored against a domain-specific rubric. Adversaries try to break the leading answer. An independent judge — different from both — rules.
The committee returns:
When consensus, adversarial survival, source grounding, and schema compliance all pass, the answer is accepted. When any one fails, the answer is surfaced for human review with the strongest dissent visible.
There is no silent acceptance.
The patterns above, applied at enterprise scale before HITS. The shape of the work — no client or system names.
Multi-stage pipeline processing high-volume public-records documents across land, court, and legal domains — text extraction, classification, field indexing, PII detection, and search-grade archival. Priority-queue work distribution, work-stealing, resume-from-stage capability. Feeds a graph database downstream.
RAG-based compliance Q&A for a specialized regulatory corpus. Serves both a conversational interface for human users and a programmatic API consumed by upstream document-processing pipelines during live ingestion. Persona-tuned prompting over a single retrieval core.
Next-generation successor to the platform above. Multi-state processing machine, tier-routed model selection (~60% per-document cost reduction), image preprocessing tuned to preserve fine detail at model-ingestion scale, and accessibility-compliant output generation.
Founded by Justin and Stacy Howard.
Senior systems engineer specializing in the ingestion-engine pattern — turning unstructured data into structured, source-attributed workflows. He's shipped this pattern five times across five distinct domains, from enterprise document intelligence at production scale to HITS's current flagship portfolio. HITS is him doing this deliberately, repeatedly, across verticals.
LinkedInStacy Howard is a co-founder of HITS. Her founder bio will be added in a near-future update.
Got a domain drowning in unstructured data? We design and build the ingestion engine that turns it into structured, source-attributed, production-grade workflows.
